Exemplarbased Inpainting of Object Removal
XU Dong-xia · Modern Electronics Technique · 2006
This paper puts forward a new method to remove the unwanted object from image,combining with the common total variation approach and image inpainting algorithm.According to the known mask,image can be separated into target region and source region.At first,the priorities of the edge pixels in target region are computed to determine the target pixel in each step; then the best matching patch is found in the source region to fill the block around the target pixel.Along with the inward propagation of priority,the object is removed and the vacancy is filled with rather satisfy pixels.Experimental results show that this method can efficiently remove big object in different shape and also function well in image gap inpainting.The result image is natural and vivid.